Config
Log for #openttd.dev on 19th September 2014:
Times are UTC Toggle Colours
00:11:14  *** Supercheese has joined #openttd.dev
00:11:14  *** ChanServ sets mode: +v Supercheese
09:28:24  *** DorpsGek changes topic to "OpenTTD Dev Channel || Latest SVN: r26850 || Logs: http://webster.openttdcoop.org/?channel=openttd.dev || Voice (talk-right) upon request via #openttd; make sure you are registered to NickServ before asking"
11:31:53  *** Supercheese is now known as Guest203
11:31:57  *** Supercheese has joined #openttd.dev
11:31:57  *** ChanServ sets mode: +v Supercheese
11:37:45  *** Guest203 has quit IRC
13:15:49  *** nickolai has quit IRC
15:11:45  *** Alberth has joined #openttd.dev
15:11:45  *** ChanServ sets mode: +v Alberth
16:38:32  *** kalenz__ has joined #openttd.dev
16:38:32  *** kalenz_ has quit IRC
16:46:38  *** frosch123 has joined #openttd.dev
16:46:38  *** ChanServ sets mode: +v frosch123
17:47:01  *** DorpsGek changes topic to "OpenTTD Dev Channel || Latest SVN: r26851 || Logs: http://webster.openttdcoop.org/?channel=openttd.dev || Voice (talk-right) upon request via #openttd; make sure you are registered to NickServ before asking"
17:53:51  *** DorpsGek changes topic to "OpenTTD Dev Channel || Latest SVN: r26852 || Logs: http://webster.openttdcoop.org/?channel=openttd.dev || Voice (talk-right) upon request via #openttd; make sure you are registered to NickServ before asking"
18:07:30  <fonsinchen> https://github.com/ulfhermann/openttd/commit/89542fac37c555d1ca9c3784a7ea2900bb233424 maybe
18:07:37  <fonsinchen> expensive, but what can we do ...
18:08:38  <fonsinchen> We have to try all cargoes to figure out what we can refit to and then we have to predict links for all refittable cargoes
18:08:46  <fonsinchen> At least I don't see an alternative.
18:19:38  <Rubidium> does autorefit require the vehicle to be empty?
18:20:07  <Rubidium> can that be used to predict that nothing is going to be changed?
18:24:25  <Rubidium> I'm not quite getting the reason to clear IN_AUTOREFIT at line 265
18:25:31  <Rubidium> besides that the comments of HandleRefit are not updated
18:25:56  <Rubidium> I've got no idea how to really test as I never used autorefit; I barely used refts ;)
18:48:52  <Supercheese> I believe wagons must be empty before they can be autorefitted
18:50:24  <Alberth> IN_AUTOREFIT  <-- comma after the last enum value
19:25:16  *** Alberth has left #openttd.dev
20:06:53  *** DorpsGek changes topic to "OpenTTD Dev Channel || Latest SVN: r26853 || Logs: http://webster.openttdcoop.org/?channel=openttd.dev || Voice (talk-right) upon request via #openttd; make sure you are registered to NickServ before asking"
20:19:20  *** Belugas has quit IRC
20:42:13  *** Belugas has joined #openttd.dev
20:42:13  *** ChanServ sets mode: +v Belugas
20:59:35  <fonsinchen> Sure they have to be empty, but this is about predicting stuff
20:59:54  <fonsinchen> you don't know if they're empty when they arrive at that stop
21:01:36  <fonsinchen> Rubidium, clearing the bit in 265 makes it do the refit theater again if the next order is another CT_AUTO_REFIT. That's intended.
21:01:50  <fonsinchen> If I didn't clear it it would only do it once for the whole order list.
21:02:51  <fonsinchen> Of course there are certain corner cases where you know that the vehicle is not empty, like with "full load all"
21:03:20  <fonsinchen> However, as refit doesn't make much sense there, it's not worth it to especially check for those.
21:05:25  <frosch123> would it be horrible to just assume that an autorefittable consist carries all refittable cargos in equal amounts? capacity 100 / 10 cargos -> 10 per cargo ?
21:06:18  <fonsinchen> Could be done, but how would that be better
21:06:24  <frosch123> or is there some way to estimate the relative cargo amounts within a network?
21:07:00  <fonsinchen> What are "relative cargo amounts within a network"?
21:07:04  <frosch123> maybe it would need less iteration
21:07:45  <fonsinchen> We'd still have to iterate all the cargoes to see which one we can refit to and the current solution keeps a cache of seen hops already
21:08:02  <fonsinchen> So, yes, marginally less iteration but I don't think it would make a difference in practice
21:08:03  <frosch123> considering all cargo sources: 10% of cargo is grain, 40% is coal, 50% is pax -> autorefittable bulk vehicles carry 20% grain, 80% coal
21:08:45  <fonsinchen> That does not tell us anything about the link we're trying to refresh
21:08:57  <fonsinchen> There might not be any coal there
21:09:53  <fonsinchen> On the other hand the vehicle could refit to full coal or full grain, so refreshing with full capacity for either is probably best.
21:11:44  <fonsinchen> https://github.com/ulfhermann/openttd/commit/a8a634d790d8a1a1ec953a8f7b56f9e654a2a685 then
21:15:55  <fonsinchen> With the "all cargoes at once" idea we'd remove iterating over orders but we'd add some iteration over the capacities and refit_capacities maps instead
21:26:21  *** frosch123 has quit IRC
22:38:32  *** nickolai has joined #openttd.dev

Powered by YARRSTE version: svn-trunk